Medical Text Classification Using Hybrid Deep Learning Models with Multihead Attention

Table 3

Analysis of results with different learning rates for the proposed quad channel hybrid model.

Learning rate	Dataset	Accuracy (%)	Precision (%)	Recall (%)	F1-score (%)
0.1	Hallmarks	59.85	61.76	60.78	61.26
	AIM	75.47	78.31	79.16	78.73
0.01	Hallmarks	73.18	71.95	72.67	72.30
	AIM	95.72	86.77	83.98	85.35
0.001	Hallmarks	72.19	70.99	72.98	71.97
	AIM	84.65	79.99	79.81	79.89
0.0001	Hallmarks	71.73	74.31	73.81	74.05
	AIM	82.36	80.91	79.73	80.31